Dementia is a contributing cause for 1 in 3 deaths, and is the only major cause of death with no effective prevention or treatment. One million Americans have advanced dementia, characterized by inability to recognize family, sparse speech, and dependency for all activities of daily living. Median survival is 1.3 years, and 6-month mortality is 25%. The final year of life wih dementia is characterized by hospitalizations for acute complications -- pneumonia and other infections, dehydration, malnutrition, falls and fractures. Acute illnesses cause symptom distress for patients, with emotional and financial strain for family caregivers. These illnesses are also associated with a high risk of death in the next year and raise ethical questions about goals of medical treatment. Palliative care consultation improves outcomes for patients with advanced cancer and critical illness, but has never been rigorously tested for advanced dementia. Dementia-specific palliative care is needed to address the unique symptoms and treatment decisions relevant to this disease. We therefore propose to develop and pilot test a model of palliative care consultation for advanced dementia patients, triggered by hospitalization for a serious acute illness. Patients will be enrolled with their family decision-makers (N=60 dyads) in a randomized feasibility trial. Intervention dyads will receive specialty palliative care consultaton during hospital admission, plus post-discharge collaborative care by their outpatient primary care provider and a palliative care nurse practitioner. Control dyads will receive usual care. Our research objective is to generate preliminary data for a large multi-site randomized controlled trial (RCT) of a model of collaborative palliative care for advanced dementia in an application to NIA (PA-13-354 Advancing the Science of Geriatric Palliative Care). Our specific aims for this application are: Aim 1: To develop a best-practice model of collaborative palliative care for advanced dementia triggered by hospital admission for serious acute illness. Aim 2: To conduct a pilot randomized trial of collaborative palliative care for advanced dementia (versus usual care) to demonstrate the feasibility of conducting a larger RCT and to estimate effect sizes to inform the larger RCT. The primary outcome will be number of hospital transfers (emergency room or hospital admission) in the 30 days post discharge. Secondary outcomes, measured at hospital discharge and 30 days follow-up, will include number of palliative care domains in the care plan, hospice referral, POLST (Physician Orders for Life- Sustaining Treatment) form completion, decision not to re-hospitalize, and burdensome treatments.